AI in HR and Recruiting
Everything around the hiring decision, deliberately not the decision.
HR is the function where the gap between what AI can do and what it should do is widest. The technology is entirely capable of ranking candidates, and doing so is where every serious legal problem in this area comes from. Senteras builds the surrounding work, which is substantial, high volume and carries none of the same exposure.
Book a free consultationWhy this is usually broken
- Automated screening and ranking carries direct EEOC exposure, and several jurisdictions now require independent bias audits of automated employment decision tools
- Removing protected attributes does not remove disparate impact, because education, geography, employment gaps and even word choice proxy for them
- HR teams answer the same policy questions hundreds of times because nobody can find the handbook
- Scheduling and coordination consume recruiter time that should go to candidate conversations
How we build it
The model does not screen or rank
This is a hard design boundary, not a setting. Ranking candidates is where the enforcement risk lives, and no efficiency gain justifies it. Everything below is chosen because it sits outside the selection decision.
Policy and handbook retrieval
Employees and managers get cited answers from your actual policy documents, with jurisdiction-specific variants handled properly rather than averaged into one answer that is wrong in three states.
Job description and requisition drafting
Draft from your own approved requisitions and levelling framework, with a check for language that narrows the applicant pool without a business reason.
Scheduling and candidate communication
Coordinate interviews and keep candidates informed, which is the single most complained-about part of most hiring processes and involves no evaluation at all.
Interview note structuring
Structure what interviewers wrote against the agreed criteria so a hiring panel compares like with like. It organises human judgement rather than substituting for it.
What changes
Figures are drawn from Senteras engagements and are illustrative of typical results. Outcomes vary by data quality, infrastructure and scope.
Common questions
Why will you not build candidate screening?
Because it is where the legal exposure is concentrated and where the technology is least trustworthy. Disparate impact does not require intent, proxy variables reproduce protected characteristics without naming them, and NYC Local Law 144 and similar rules now require independent bias audits. If you want it built, we would rather tell you plainly that we think it is a bad trade than build it and be quiet about the risk.
What if we keep a human in the loop on screening?
A human approving a model-generated ranking is usually a rubber stamp in practice, and regulators have said as much. The ranking is the decision. That is different from a human structuring their own notes against agreed criteria, which is what we do build.
Is policy retrieval really worth a project?
For most HR teams it is the highest-volume repetitive work they do, and the answers are already written down. It is unglamorous, it carries no legal exposure, and it returns hours per week immediately.
Can it handle multi-state policy differences?
It has to. Leave entitlements, pay transparency and notice requirements differ by state, and a system that averages them produces an answer that is confidently wrong in most of them. Jurisdiction is part of the retrieval, not a caveat in the response.
Where we deploy this
Professional Services
Turn your firm's accumulated knowledge into a scalable competitive advantage.
Technology & Software
Ship faster, support better, and build AI-native products your competitors can't match.
Healthcare & Life Sciences
Reduce administrative burden and surface clinical intelligence, with privacy built in.
Manufacturing & Supply Chain
Reduce downtime, improve quality, and build a more resilient supply chain with AI.
Retail & Consumer Goods
Predict what customers want before they ask, and stock exactly what you need.
The services behind it
Local & On-Prem LLM Deployment
The most powerful AI models, running entirely on your hardware.
Custom AI Agents & Automation
AI that doesn't just answer questions. It gets things done.
Training & Change Management
Technology alone doesn't transform organizations, people do.
Find out what this looks like for your organisation
A 30-minute call. We will tell you plainly whether AI is the right tool for the problem you have, including when it is not.
Book a free consultation